Metrics for Bio-Inspiration Based on Taxonomies
Total Page:16
File Type:pdf, Size:1020Kb
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED21 16-20 AUGUST 2021, GOTHENBURG, SWEDEN METRICS FOR BIO-INSPIRATION BASED ON TAXONOMIES Willocx, Mart (1); Duflou, Joost R. (1,2) 1: KU Leuven; 2: Member of Flanders Make ABSTRACT Using bio-inspiration allows engineers to use the knowledge implicitly built up by natural evolution. Current tools for providing engineers with bio-inspiration yield many biological working principles. Starting from the Linnaean taxonomy, which can be seen as a design revision history, this work proposes metrics for a working principle based on the observations of that working principle in different organisms. A first metric measures the reinforcement of a working principle via the number of observations (publications/submissions to a database) made by biologists. Furthermore, biological strategies that evolve independently and use the same working principle might be more resilient and globally applicable, prompting the proposal of a metric measuring the spread in the taxonomy. Finally, bio-novelty measures the biological novelty, inversely related to the biological diversity employing the working principle. To illustrate the use of the metrics, they are applied to the working principles identified in the ‘temporary attachment’ category of AskNature. Keywords: Bio-inspired design / biomimetics, Computational design methods, Early design phases, Bio-inspiration Contact: Willocx, Mart KU Leuven Mechanical engineering Belgium [email protected] Cite this article: Willocx, M., Duflou, J. R. (2021) ‘Metrics for Bio-Inspiration Based on Taxonomies’, in Proceedings of theICED21 International Conference on Engineering Design (ICED21), Gothenburg, Sweden, 16-20 August 2021. DOI:10.1017/1 pds.2021.470 ICED21 2087 1 INTRODUCTION Even before Leonardo da Vinci’s flying machines, humans have been learning from nature. Evolution has already solved many problems faced by engineers today. Due to the iterative evolution process, strategies found in nature have proven their performance (Chirazi et al., 2019). Bio-inspired design aims to use this knowledge by transferring it to engineering. Using an analogical transfer, the working principles of these strategies are applied in engineering designs (Hashemi Farzaneh and Lindemann, 2019). For example hook and loop fasteners were created after extracting the working principle of plant burrs and their ability to attach to fur (Yen et al., 2014). A major issue of the current tools created for supporting engineers in finding relevant bio-inspiration is the number of biological strategies that are retrieved (Kaiser et al., 2013; Vandevenne et al., 2015). To alleviate this information overload, Vandevenne et al. proposed to cluster the biological strategies based on the working principle employed by the organism (Vandevenne et al., 2015). The relevance of these working principles depends on the qualities desired by the searching engineer: they might be more interested in a proven working principle to create a robust design or desire a more novel working principle to support innovation. Reranking the retrieved bio-inspiration to take these desires into account might decrease the number of working principles that have to be evaluated during the selection & analysis phase. This work argues that metrics based on the taxonomy of the identified focus organisms give an indication of the performance, biological novelty and existing biological interest in the identified working principles. During a systematic search for bio-inspiration, the designer can use these metrics to quickly evaluate the possible usefulness of the retrieved working principles in view of the objectives of the design assignment. As these metrics can be automatically calculated, they can be integrated in a design support tool to rank the resulting clusters of biological documents based on the desired qualities of the working principles. The next sections first introduce the relevant literature for systematically finding bio-inspiration, supporting filtering and selecting bio-inspiration and different forms of biological taxonomies. Second, existing methods for calculating distances in these taxonomic trees are reviewed. Third, three metrics are proposed to rank the working principle clusters based on the occurrences of their focus organisms in a taxonomy. Finally, to illustrate the use of the metrics, they are applied to the working principles identified in the case study from (Willocx et al., 2020) on ‘temporary adhesion’. 1.1 Systematically finding bio-inspiration To integrate bio-inspiration into systematic engineering design (engineering pull of bio-inspiration), a systematic design process has been proposed as presented in figure 1. The search for bio-inspiration starts from analysing the problem and formulating a problem description, which is used in a search operation. From the resulting biological information, the most relevant strategies are selected before they are transferred to the engineering domain (Vandevenne et al., 2015). Problem Selection & Knowledge Search formulation analysis transfer Figure 1: The phases of the systematic bio-inspired design process as identified by Vandevenne et al. (2015). Already several search tools have been developed which can be roughly categorized as three different approaches: direct consultation of a biologist, using a specially prepared database and using natural language biological documents for identifying bio-inspiration (Willocx et al., 2020). A major drawback of consulting an expert biologist is the limit of his or her current knowledge and the possible bias to his/her own research domain (Shu et al., 2011; Graeff et al., 2020). The major issue with database approaches is the enormous effort that is required for the population of the database with biological strategies (Vattam et al., 2011; Graeff et al., 2019a). The natural language search tools result in a large quantity of potentially relevant biological documents, which makes the selection of relevant documents time-consuming (Kaiser et al., 2013; Willocx et al., 2020). Furthermore, as shown in (Willocx et al., 2020), due to the focus on the organism performing the strategy, many parallel strategies employing the same principle are also submitted in a database like AskNature. 2088 ICED21 To counter this information overload, Vandevenne et al. proposed to group the retrieved documents based on the focus organism of the biological text, aiming to cluster similar biological strategies and reduce the amount of information the engineer has to navigate (Vandevenne et al., 2015). However, this does not take convergent evolution into account, where a similar working principle arises independently in two different organisms (Losos, 2011). Vandevenne et al. propose to do this clustering based on the enabling function of the biological strategy for which a method to automatically extract these is available in (Cheong and Shu, 2014) (Vandevenne et al., 2015). 1.2 Related tools and methods for filtering and selecting bio-inspiration Most engineers do not have a background in biology, making filtering and understanding the retrieved biological documents a time-consuming task which can result in fixation on one biological strategy, an incorrect analogical transfer or using ‘off-the-shelve’ biological solutions (Helms et al., 2009; Vattam and Goel, 2011; Graeff et al., 2019b). These pitfalls can be alleviated by including a biologist in the design team. This biologist performs a constant pre-evaluation of the biological strategies encountered, keeping only the most relevant strategies for consideration (Lenau et al., 2011). To be able to recall a large variety of relevant biological strategies, this biologist needs to have a wide basis of biology, which is a rare profile (Graeff et al., 2019a). Tools and methods supporting filtering and selecting biological inspiration rely on having the engineer understand the retrieved strategies and manually select the most relevant for their problem (Lenau et al., 2018). Lenau et al. highlight this for the ISO approach (ISO/TC 266, 2011), the BioCards approach (Lenau, 2017) and the generic bio-inspired design method developed in (Fayemi, 2014). This reliance on the engineer to understand the biological strategy does not alleviate the time-consuming nature of filtering the retrieved bio-inspiration. 1.3 Using taxonomies as a design history Evolution is an iterative process, resulting in the organisms currently roaming the earth. The genealogical history of organisms is written inside their genes. A Linnaean taxonomy aims to organize organisms with similar properties (e.g. morphological, genetic…) into coherent units, so-called taxa. These taxa are organized in a hierarchical classification with several ranks (Ruggiero et al., 2015), capturing an evolutionary history of the organisms involved. With modern molecular sequencing, a phylogenetic tree can be built based on the genetics of the organisms involved, capturing the genetic history of the organism (Woese, 2000). The genetic distance between organisms is based on the number of mutations and evolutionary events since their divergence (Nei, 2001). Similar adaptations to taxonomically unrelated organisms can occur when the organisms occupy similar environmental contexts. This convergent evolution of the working principle is the result of natural selection of a good strategy (Losos, 2011). For example: for providing temporary attachment to diverse surfaces, AskNature contains strategies from geckos (Autumn et al., 2006),